SWEETAPPLE: POSSUM DIET IN 157

Possum (Trichosurus vulpecula) diet in a mast and non-mast seed year in a New Zealand forest

P. J. Sweetapple Landcare Research, P.O. Box 69, Lincoln 8152, New Zealand (E-mail: [email protected])

______Abstract: The annual diet of possums (Trichosurus vulpecula) during both a beech (Nothofagus) mast fruiting year and a non-mast year in the simple beech of the North Branch of the Hurunui Catchment, eastern South Island, New Zealand, was determined by sorting the contents of 270 possum stomachs, collected between December 1999 and December 2001. Beech flowers and seeds contributed 46.1% to annual diet during the mast year, but were not eaten during the non-mast year. Beech foliage and bark made up 13.2% and 45.0% of annual diet in the mast and non-mast years, respectively. Fungi, herbs and grasses together comprised 23.1% and 42.4% of the annual diet in the mast and non-mast years, respectively. Diet varied with altitude and distance down the valley, principally in the relative contributions of foods from the three beech species present, which reflected local canopy composition. Other local food sources such as valley-floor herbs and grasses also contributed significantly to local diet. Although heavily reliant on beech species, possums are unlikely to have a significant impact on beech flower and seed production, or on the health of beech because of possums’ low abundance and the large beech biomass at this site. Recognised possum-preferred foods such as Aristotelia serrata, Fuchsia excorticata, Hoheria glabrata, simplex, Elaeocarpus hookerianus and mistletoes (Peraxilla tetrapetela and Alepis flavida) were all strongly preferred by possums but, because of their scarcity in the study area, were only eaten in small quantities. ______Keywords: diet; mast fruiting; possum; simple Nothofagus forest; Trichosurus vulpecula.

Introduction threaten indigenous forest ecosystems through their herbivory and predation [see Payton (2000) and Sadleir Beech (Nothofagus) species exhibit mast seeding with (2000) for reviews]. In many habitats possums eat large quantities of seed (> 4000.m-2) produced at large quantities of and seed when these are intervals averaging c. 5 years, with little or no seed available (Nugent et al., 2000). Periods of heavy fruit produced in many of the intervening years (Wardle, utilisation have been linked with enhanced breeding 1984). Partial mast years (500–4000 seeds.m-2) are success in possums (Bell, 1981), and possums have sometimes observed between full mast years (Wardle, been shown to suppress fruit production in some 1984). Beech masting is a significant ecological event species (Cowan, 1990a; Cowan and Waddington, in beech forest: heavy flower and seed production 1990). Beech masting may, therefore, affect possum triggers eruptions in forest-litter arthropod populations diet and fecundity, and possum feeding may affect (Alley et al., 2001) and increased breeding activity in beech seed production, but these potential interactions some indigenous birds such yellow-crowned parakeet have not been quantified. Furthermore, there are no (Cyanoramphus auriceps) and kaka (Nestor published data on possum diet from the floristically meridionalis) (Elliott et al., 1996; Wilson et al., 1988). simple beech forests that dominate the indigenous Beech mast events also trigger perturbations in the forests east of the Main Divide of the Southern Alps in densities of introduced mammalian pests, including the South Island and the axial ranges of the North house mice (Mus musculus) (King, 1982, 1983; Island (Wardle, 1984). Possum diet has been described Murphy, 1992; Fitzgerald et al., 1996; Alley et al., from some western South Island beech forests (Owen 2001) and stoats (Mustela erminea) (King, 1983). and Norton, 1985; Cochrane et al., 2003). Elevated stoat densities often result in increased The New Zealand Department of Conservation predation on indigenous birds (Murphy and Dowding, (DOC), Canterbury Conservancy, contracted Landcare 1995; Elliott, 1996; Elliott et al., 1996; O’Donnell and Research, Lincoln, New Zealand, to determine possum Phillipson, 1996; Wilson et al., 1998). diet from stomachs collected from the North Branch Possums (Trichosurus vulpecula) are one of New Hurunui Catchment, eastern South Island in 1999– Zealand’s most pervasive mammalian pests, and 2001. Seedfall trays in the adjacent South Branch

New Zealand Journal of Ecology (2003) 27(2): 157-167 ©New Zealand Ecological Society 158 NEW ZEALAND JOURNAL OF ECOLOGY, VOL. 27, NO. 2, 2003

Hurunui Catchment caught 160.7 kg.ha-1 of beech throughout, but dominating the canopy at high altitudes seed from March to May 2000 (A. Grant, Department and in wetter western parts of the study area. Mountain of Conservation, Christchurch, N.Z., unpubl. data). beech (N. solandri var. cliffortioides) is common in the This equates to c. 4000 seeds.m-2 (Wardle, 1984: p. drier eastern parts of the study area, particularly at high 256) indicating that the first year of the study was a full altitudes. In total, beech species comprise 82% of the beech-mast year. This estimate is likely to underestimate estimated total biomass in the study area: 21%, total seedfall as small quantities of seed continue to fall 27% and 34% for mountain, red and silver beech, during winter and, for silver beech, spring (Wardle, respectively [data derived from the National Vegetation 1984). The following year, 2001, was a non-mast year Survey (NVS) databank]. The sparse non-beech with just 0.5 kg.ha-1 of beech seedfall in March to May. components of these forests are dominated by small- The diet study provided the opportunity to compare leaved Coprosma species, Griselinia littoralis, possum diet in a beech mast and non-mast year, as well Phyllocladus alpinus and Myrsine divaricata. Possum- as to document possum diet from a simple eastern preferred tree species are uncommon or rare, ranging beech forest and assess variation in diet with altitude, from about 0.5% [Raukaua simplex (Mitchell et al., aspect and geographic location. The results of this 1997), Hoheria glabrata] to less than 0.05% (Aristotelia work are presented in this paper. serrata, Fuchsia excorticata) of estimated plant biomass (NVS databank). Subalpine and alpine shrublands, grasslands and herbfields are present above Methods the upper limit of forest. Possums have been present in the study area since Study area at least 1963 (Cowan, 1990b) and leg-hold trap-catch The study area consisted of the forested parts of the rates of 8–16 possums per 100 trap-nights during the upper North Branch Hurunui Catchment, North study period (A. Grant, Department of Conservation, Christchurch, N.Z., unpubl. data) indicate a low-density Canterbury, New Zealand, from Harper Pass on the -1 Main Divide of the Southern Alps to Mackenzie Stream, possum population (c.1–2 possums.ha : B. Warburton, 14 km east of the Main Divide (Fig. 1). The valley floor Landcare Research, Lincoln, N.Z., pers. comm.). is characterised by broad river flats at 600–700 m a.s.l., much of which are open grasslands dominated by Possum stomach collection exotic species. The valley sides are almost entirely Stomachs were taken from possums caught on the forested by beech up to c. 1300 m a.s.l. Red beech same six transects during each sampling trip. Transects (Nothofagus fusca) occurs on the lower and mid- were subjectively located to sample the north and altitudes slopes, with silver beech (N. menziesii) present south sides of the valley, and the western, middle and

River/streams N

Forest margins Mackenzie Stream Sampling transects

C Cyanide-poisoned (May) T Trapped (March)

0 1 2 C

km Lake Sumner 5 km

Landslip Stream

Blue Stream

T C T T T T T C

C

a T m e C r T o C n

T S

t

r

e

a C m

Harper Pass Figure 1. The North Branch Hurunui study Upper Valley Mid-Valley Lower Valley Transects Transects Transects area and location of sampling transects. SWEETAPPLE: POSSUM DIET IN BEECH FOREST 159

eastern parts of the study area (Fig. 1). They ran from The percent dry weight for each food item identified the forest-grassland margin on the valley floor to the was calculated for each stomach. The mean percent upper altitudinal limit of tall forest. Between 13 and 25 dry weight (all stomachs) for each collection was then sampling stations were established at approximately calculated for each food item, and the annual means 25-m vertical intervals along each transect. Cyanide were calculated by averaging the collection means. baits (Feratox®, Feral Control, Auckland) were laid in Differences between years in the contribution to diet of bait stations at each sampling station for about 4 days food items that comprised more than 0.5% of annual (range: 2–5 days), in December, February/March, May/ diet in either year were tested using Kruskal-Wallis June, and September, in each of the two years. A total tests. Because the sampling regime was balanced (strata of 360 possum stomachs were collected during the were sampled in proportion to their size, with the same study. Of these, 270 were sorted and analysed (30 in lines sampled each season, and the same seasons February/March in both years and 35 for all other sampled each year) there was no need to account for collections). When more than 35 stomachs were stratum and seasonal effects to obtain an unbiased collected in the same sampling period, those sorted had estimate of mean annual possum diet. Statistical been randomly selected from the whole sample. analyses were performed using the Systat (version 6) Each possum sampled was allocated to three strata: statistical package. altitude, geographic location, and the side of the valley (aspect). These strata were defined as low Preference and utilisation indices (600–800 m), mid (800–1100 m), and high (1100– Preference ratings express the relationship between 1300 m) for ‘altitude’; upper, mid, and lower valley for proportional use of a plant and its proportional ‘geographic location’ (Fig. 1); and north and south abundance. They were calculated for foliage and bark side for ‘side of valley’. The sampling strategy resulted (combined) of forest plant species identified in stomach in varying numbers of stomachs analysed per stratum, samples. Relative foliage biomass can be quickly and reflecting the relative availability of each stratum and relatively accurately estimated from cover scores (Spurr relative possum density in each. The resulting total diet and Warburton, 1991). Therefore, plant availability data therefore are weighted toward those strata where was estimated from cover scores on RECCE plots most possums were caught and, hence, are an unbiased (Allen, 1992), using similar methods to Owen and measure of the mean per-capita diet. Norton (1995). Sixty five RECCE plots, one at every Relative possum density was compared amoung second sampling station along all six transects, were strata by calculating the capture rate (expressed as a measured in 2000/2001 (NVS databank). On these percentage of bait nights) for each stratum on each line plots, plant species present were given one of six cover every sampling period. However, data for low and scores (<1%, 1–5%, 5–25%, 25–50%, 50–75%, >75% high altitudes in the upper-valley stratum were excluded of the plot area) in each of seven tiers (0–0.3, 0.3–2.0, because there were less than four bait stations in each 2.0–5.0, 5.0–12.0, 12.0–20.0 m above ground, stratum. Mean capture rates and 95% confidence emergent, and epiphyte). Cover scores were then intervals were calculated for each stratum. weighted by their mid points (e.g. the 5–25% cover class was weighted by 15) and by the vertical intervals Diet analysis of the tiers in which they were recorded (e.g. the 2.0– Stomachs with layered contents were sorted using the 5.0-m tier was weighted by 3.0, with emergents ‘layer separation’ technique (Sweetapple and Nugent, weighted by 8.0, the same as for canopy trees, and 1998), where discrete layers of single food items epiphytes weighted by 0.5). Weighted cover scores within the stomach are separated, oven dried at 70°C were then summed by species to give estimates of for 24 hours, and weighed. For the 1.8 % of whole or relative availability for each plot. Mean proportional part stomachs without discernable food layers, contents availability was then determined for each species over were sorted by sieving through a 2-mm mesh, then all plots by dividing the total availability of each identifying, drying, and weighing 100 fragments species by the total availability of all species. systematically selected from the material retained in The preference index (PI) used was the forage the sieve (Sweetapple and Nugent, 1998). ratio (Loehle and Rittenhouse, 1982) and is calculated Stomach contents were identified by species and as: food type (foliage, fruit, flower, bark, and seed) for woody species and ferns, but herbaceous material from the valley-floor grasslands was pooled, as was herbaceous material from the forest floor, and all where % diet = the percentage of total foliar diet. fungi. Beech seed was identified to species by seed Preference index values below 0.5 indicate low size, seed wing characteristics, and beech foliage traces preferences and PI values above 2.0 indicate high or small foliage layers found within the seed layers. preferences. For forest recorded in possum diet 160 NEW ZEALAND JOURNAL OF ECOLOGY, VOL. 27, NO. 2, 2003

but not on RECCE plots, a minimum PI was calculated Table 1. Possum sampling effort, possum catch-rate (possums by using the minimum possible abundance score of caught/100 bait-nights ± 95% confidence limits) and the number of stomach samples analysed for each stratum in the North 0.15 (one occurrence in the 0–0.3-m tier with a cover Branch Hurunui Catchment. score of 1). ______If possum densities vary markedly between Stratum Bait Bait- Possums Catch- Stomachs stations nights caught rate analysed statistical strata, then PIs are of limited use for ______comparing relative browse pressure between strata. Altitude Therefore, a utilisation index (UI) was calculated for Low (600–800 m) 33 1169 166 14 ± 3 126 each stratum for high preference foods as follows: Mid (800–1100 m) 74 2554 163 6 ± 2 119 High (1100–1300 m) 21 738 33 4 ± 2 25 Side of valley North 65 2227 156 7 ± 2 131 South 63 2234 206 9 ± 2 139 Geographic location where Da = possum capture rate in stratum a, % dieta = Lower valley 50 1825 152 8 ± 1 112 percentage of total diet in stratum a, and % abundancea = percentage of abundance in stratum a. Mid valley 50 1725 140 8 ± 2 103 Upper valley 28 911 70 8 ± 4 55 Total 128 4461 362 8 ± 1 270 ______

Table 2. Mean percent dry weight of food items in a beech-mast (1999–2000) and non-mast (2001) year. Kruskal-Wallis test statistics (H) and P values for the comparison of importance of each diet item between the two years are shown. Significant results at the 95% confidence level are indicated (*). Preference indices (PI) are given for foliage categories (includes bark for beech species) for foods occurring within the forest. Minor foods (< 0.5% in both years) are listed at the bottom of the table together with their percent contribution to total diet (both years combined) and their preference indices (in brackets). ______Food item Mast year Non-mast HPPI (%) year (%) ______Nothofagus pooled by food type Beech foliage 10.6 35.6 51.7 < 0.001* Beech seed 37.1 0.0 109.5 < 0.001* Beech bark 2.6 9.4 4.8 0.028* Beech flowers 9.0 0.0 18.5 < 0.001* Nothofagus pooled by species Red beech 23.2 30.7 7.0 0.008* 1.7 Mountain beech 25.4 12.5 6.5 0.011* 0.9 Silver beech 10.6 1.8 7.1 0.008* 0.1 Total Nothofagus species 59.3 45.0 11.2 0.001* Other foods Fungi 8.6 18.3 2.9 0.090 - Valley-floor herbs and grasses 10.3 17.7 3.6 0.051 - Forest-floor herbs 5.3 6.7 0.1 0.777 6.1 Raukaua simplex foliage 2.4 2.2 0.6 0.439 5.1 Muehlenbeckia axillaris foliage 0.9 2.6 2.3 0.138 - Hoheria glabrata foliage 1.4 1.8 0.8 0.373 18.0 Polystichum vestitum 2.2 0.3 0.9 0.351 13.5 Small-leaved Coprosma foliage 1.0 0.9 < 0.1 0.878 0.4 Coprosma foetidissima foliage 0.6 0.8 0.3 0.582 1.13 Aristotelia serrata foliage 0.1 1.0 2.9 0.091 > 120.0 Rubus spp. foliage 0.9 0.2 0.9 0.335 4.4 Podocarpus nivalis fruit 0.9 0.0 2.2 0.155 - Blechnum spp. 0.3 0.5 0.1 0.751 10.0 Podocarpus nivalis foliage 0.7 < 0.1 3.5 0.063 1.6 Mistletoe spp. foliage 0.5 0.0 3.9 0.048* > 60.0 Other items pooled 4.6 2.2 - ______Foods < 0.5% in both years: moss/litter/soil, 0.26; Rubus spp. flowers, 0.25; unidentified bark and wood, 0.25; Gaultheria depressa fruit, 0.21; Fuchsia excorticata foliage, 0.18, (46); unidentified foliage, 0.09; Elaeocarpus hookerianus foliage, 0.08, (> 16); crassifolius foliage, 0.07, (0.2); Gaultheria antipoda/depressa foliage, 0.04, (0.3); Myrsine divaricata foliage, 0.03, (0.1); Aristotelia fruticosa foliage, 0.02, (> 4); Invertebrates, 0.13; Hebe species foliage, 0.01, (< 0.01); Muehlenbeckia axillaris fruit, 0.01; Hypolepis species, < 0.01, (0.1); Raukaua anomalus foliage, < 0.01, (0.5). SWEETAPPLE: POSSUM DIET IN BEECH FOREST 161

Results Beech seeds & flowers 60 Beech leaves & bark Possum density and distribution The mean possum capture rate throughout the study was 8 ± 1 (95% confidence limits) possums per 100 40 cyanide bait-nights. The capture rate was identical in Fungi & herbs both years indicating that repeated sampling of the 20

transects did not reduce possum abundance. Possums Percent of diet were evenly distributed along the length of the catchment and on both its north and south sides, but were more abundant on the lower forested slopes 0 Dec Feb May Sep Dec Mar June Sep Dec (Table 1). Month mast year non-mast year Possum diet Foods from beech species (pooled) were the most Figure 2. Contribution to possum diet of beech seeds and flowers (circles), beech leaves and bark (squares) and fungi and significant contributors to possum diet, totalling 59% herbs (triangles) during each collection period, in the North and 45% of annual diet during the mast and non-mast Branch Hurunui. Mast year 1999–2000, non-mast year 2001. years, respectively (Table 2). Substantial quantities of all three beech species present were eaten, with red beech, followed by mountain beech, the most important beech food species, while foliage then seed were the In total there was a significant but modest decline most important beech food types (Table 2). in consumption of all beech foods from the mast to the Food type composition of the beech component of non-mast year (Table 2). This difference was possibly annual diet varied markedly between the two years made up for by increases in fungi and valley-floor (Table 2, Fig. 2). Seeds and flowers contributed 46% herbs, although statistical tests were non-significant to total diet during the mast year but were not eaten (Table 2). These two foods were the next most important during the non-mast year (Table 2). Beech seed was food items after beech in both years, each contributing almost exclusively foraged from within beech canopies between 8.6 and 18.0% to annual diet (Table 2). in all seasons, as seed food layers often contained Fungal material mainly consisted of basidiomycete traces of green beech foliage but no litter, fungi or sporocarps (mushrooms), and was eaten year round herbs as would be expected if seed was eaten while but particularly from February to June (Table 3). possums foraged on the ground. Possums were feeding Valley-floor herbs and grasses were also eaten year heavily on beech seed (45% of diet) before it was ripe round, with consumption greatest in September and in February 2000, and were still eating large quantities December (Table 3). This food group was dominated of it (37% of diet) in September 2000 (Fig. 2). Beech by clover (Trifolium species) and fine grasses flowers were only eaten in December 1999, they (Gramineae). They contributed 24.1% of annual diet appeared to be mainly the anthers of male flowers, and at low altitudes but < 6.0% at higher altitudes. One were largely from mountain beech at high altitude possum travelled over 1 km and at least 550 m vertically sites. after feeding on the valley-floor grasslands. The absence of beech seed and flowers in the non- Forest-floor herbs were the only other food to mast year was largely compensated for by a greater exceed 3.0% of annual diet in either year (Table 2). than threefold increase in the consumption of beech Prominent components of this group included Nertera foliage and bark in that year (Table 2, Fig. 2). Red villosa, Viola cunninghamii, Gunnera monoica and beech dominated the beech foliage food group in the Pratia angulata. non-mast year, particularly in early summer (December) Recognised possum-preferred forest species such when it contributed 46% to possum diet (Table 3) and as haumakaroa (Raukaua simplex), ribbonwood was comprised exclusively of soft new foliage. Beech (Hoheria glabrata), wineberry (Aristotelia serrata), bark eaten was thin, often green, and included some mistletoe (Alepis flavida and/or Peraxilla tetrapetela), woody material, clearly sourced from small-diameter fuchsia (Fuchsia excorticata) and pokaka (Elaeocarpus branches. Bark was sometimes eaten in large quantities. hookerianus) were all small contributors to total diet in It made up more than 50% of contents in 7% of both years. However, they were all strongly sought stomachs. Although beech foliage and bark were after, having high preference indices (Table 2), which important components of possum diet, the large biomass indicates that low levels of consumption were a of these species in the study area meant these foods did consequence of low abundance of these species in not gain high preference indices (Table 2). these forests, some of which (e.g. mistletoe, pokaka, 162 NEW ZEALAND JOURNAL OF ECOLOGY, VOL. 27, NO. 2, 2003

Table 3. Mean seasonal and annual possum diet (December 1999–December 2001) in the North Branch Hurunui Catchment. ______Food Item Feb–March May–June September December Annual ______Main food groups Foliage (woody species) 18.5 30.0 35.4 47.0 32.6 Foliage (herbaceous species) 19.9 8.9 34.0 21.6 21.1 Fruit and seeds 26.6 33.0 18.3 < 0.1 19.5 Fungi 20.5 21.5 2.6 10.0 13.7 Bark 10.7 5.6 8.4 < 0.1 6.2 Flowers 0.0 0.0 0.0 19.2 4.8 Ferns 3.1 0.7 0.7 2.0 1.6 Other foods 0.7 0.3 0.6 0.2 0.5 Individual food items Nothofagus fusca foliage 2.6 17.8 11.1 35.8 16.8 Valley floor herbs and grasses 10.8 6.0 22.9 18.2 14.5 Fungi 20.5 21.5 2.6 9.9 13.6 Nothofagus fusca seed 5.4 8.0 14.9 0.0 7.1 Nothofagus solandri var. cliffortioides seed 11.7 13.4 3.2 0.0 7.1 Forest floor herbs 8.3 2.9 9.2 3.5 6.0 Nothofagus solandri var. cliffortioides foliage 2.3 8.1 9.6 0.1 5.0 Nothofagus menziesii seed 5.9 11.6 0.2 0.0 4.4 Nothofagus solandri var. cliffortioides flowers 0.0 0.0 0.0 16.9 4.2 Nothofagus fusca bark 2.5 2.4 6.4 < 0.1 2.8 Nothofagus solandri var. cliffortioides bark 7.3 2.3 0.5 0.0 2.5 Raukaua simplex foliage 2.1 1.5 5.4 < 0.1 2.3 Muehlenbeckia axillaris foliage 3.5 0.1 0.0 3.3 1.7 Hoheria glabrata foliage 3.7 0.1 0.0 2.6 1.6 Polystichum vestitum 3.1 0.7 0.7 0.6 1.3 Nothofagus menziesii foliage < 0.1 0.7 4.0 0.0 1.2 Small leaved Coprosma spp. foliage 1.4 < 0.1 0.9 1.6 1.0 Coprosma foetidissima foliage 1.4 0.4 0.4 0.6 0.7 Aristotelia serrata foliage 0.4 0.0 1.8 0.0 0.6 Nothofagus fusca flowers 0.0 0.0 0.0 2.2 0.6 Nothofagus menziesii bark 0.0 0.7 1.5 0.0 0.6 Rubus spp. foliage 0.4 0.0 0.0 1.9 0.6 Podocarpus nivalis fruit 1.8 0.0 0.0 0.0 0.5 Blechnum spp. 0.0 0.0 0.0 1.5 0.4 Podocarpus nivalis foliage 0.1 0.3 1.0 0.0 0.4 Mistletoe foliage < 0.1 1.0 < 0.1 0.0 0.3 Rubus spp. flowers 0.0 0.0 0.0 1.0 0.3 Other foods 4.9 0.5 3.7 0.3 2.7 ______wineberry) were not recorded on any of the RECCE but these foods had low to moderate preference indices plots. (Table 2). With the exception of mistletoe, all minor foods Invertebrates made up just 0.1% of total diet, (maximum annual consumption = 0.5–3.0%) were although this included several hundred lepidopteran eaten in similar quantities in both years (Table 2). larvae, mainly eaten on the valley-floor grasslands in Mistletoe foliage was recorded from four stomachs September 2001. during the mast year. For each record, mistletoe was Dracophyllum longifolium, Griselinia littoralis, closely associated with beech seed food layers. Olearia lacunosa, Phyllocladus alpinus and Mistletoe was not recorded in possum stomachs during Pseudopanax crassifolius each accounted for ≥ 0.5% the non-mast year. of total plant biomass within the study area (NVS Other minor foods with high preference indices databank) but were not recorded in possum diet. were the ferns Polystichum vestitum and Blechnum species (B. procerum and B. penna-marina pooled), Comparison of utilisation indices across strata and foliage of Rubus species (Table 2). Foliage from Utilisation indices for high preference foods ranged small-leaved Coprosma species and Coprosma from very high (> 100) to low (< 10) across different foetidissima was also regularly eaten in small quantities strata for different foods. These differences (Table 4) SWEETAPPLE: POSSUM DIET IN BEECH FOREST 163

Table 4. Utilisation indices for the main high preference food groups in the eight altitude and valley zone strata. Missing data indicate that the food item was not recorded on RECCE plots in that strata. ______Altitude strata Location within valley Side of valley

Food item low mid- high upper mid- lower north south ______Forest-floor herbs 32 34 30 28 32 41 32 46 Raukaua simplex foliage 42 10 7 16 13 22 23 12 Hoheria glabrata foliage 0 66 – 38 429 – 49 837 Polystichum vestitum 34 34 144 33 23 108 39 87 Rubus spp. foliage 19 13 - 1 208 11 15 60 Blechnum spp. 33 40 7 4 91 0 28 28 ______

did not correspond to possum-density patterns across possums during the mast year (Table 2). The resultant the study area. Very high utilisation was recorded at high preference index of 10.2 (59%/5.7%) clearly mid-valley sites for ribbonwood, Rubus species and demonstrates that possums actively foraged for beech Blechnum species, at south-side sites for ribbonwood seed during the mast year. and at high altitude sites for Polystichum vestitum, yet Assuming there is a mean possum density of about the utilisation of these foods was usually low or 1–2.ha-1 in the study area and that wild possums moderate (10–50) in other strata (Table 4). Utilisation consume about 160 g (dry weight) of food per night of ribbonwood, Rubus species and Polystichum (Fitzgerald et al., 1981), possums ate c. 22–44 kg vestitum was markedly greater on the south (north ha-1 of beech seed during the mast year. This equates facing) side than on the north side of the study area to about 14–27% of the assumed March–May seedfall (Table 4). This reflects the relative abundance of these in the study area, and indicates that possums probably species on each side of the catchment, with all three caused only a modest reduction in total seedfall. The least abundant on the south side (NVS databank). proportion of the total seed crop eaten by possums is likely to be less than this because some beech seed continues to fall during winter and spring (Wardle, Discussion 1984). Flowering in beech starts at low altitudes in mid- Beech flowers and seed in possum diet spring and progresses up-slope at about 300 m per The mast seeding of beech during the study had a month, and lasts about 2 weeks at any one location marked effect on the feeding activity of possums in the (Wardle, 1984). The observation that most possums study area. Beech seeds and flowers featured that ate beech flowers were caught at high altitudes prominently in possum diet during the beech-mast probably reflects this variable timing of flowering at year, but were not eaten at all during the second year of different altitudes. Flowering may have largely finished the study. This was not unexpected, as possums at lower altitudes when the December 1999 sample elsewhere feed heavily on most and many flowers was taken. Possum feeding on beech flowers is likely when they are available (Nugent et al., 2000), and to have had a negligible affect on seed set given that possums have been recorded travelling several hundred flowering was probably prolific and of short duration, metres from their usual home ranges to feed on beech that possums apparently targeted male flowers, and flowers (Ward, 1978). that beech flowers are wind pollinated (Wardle, 1984). An approximate beech seed preference index for The consumption of beech seeds and, in particular, the mast year can be calculated from data presented flowers highlights the plasticity of possum feeding here and in the literature as follows. Annual foliage behaviour, with large and rapid shifts in feeding activity production for montane red-silver beech forest is enabling possums to utilise new and ephemeral food c. 2500kg.ha-1 (Sweetapple and Fraser, 1992). The sources. biomass of beech flowers produced is similar to the The heavy and prolonged (at least 8 months) subsequent seed crop (161kg.ha-1 in 2000) (Alley et consumption of beech seed by possums during a beech al., 2001), so c. 2820kg.ha-1 of foliage, flowers and mast year potentially adds another dimension to the seed were produced by woody forest vegetation in the perturbations in mammalian pest numbers in beech study area during the mast year, of which 5.7% was forests known to occur during and after beech mast beech seed. Beech seed made up 59% of all foods from years (e.g. King, 1983). In years when quality food woody forest vegetation (excluding bark), eaten by resources, such as fruits, are abundant, adult female 164 NEW ZEALAND JOURNAL OF ECOLOGY, VOL. 27, NO. 2, 2003

possum body condition (weight) is higher than average, study area. Unless the nutritional properties of beech promoting earlier breeding, higher pouch young species differ markedly from area to area, the results of survival (Bell, 1981), and spring breeding (Cowan, the current and other diet studies demonstrate that 1990b). This may raise possum densities above the beech foliage and bark are in fact palatable, and despite long-term average, which can result in increased being of low preference to possums (they are not eaten browsing pressure on, or even death of, preferred tree at all in some beech forests) these foods sometimes species (e.g. Cowan et al., 1997). Masting in beech contribute significantly to annual possum diet in the species, particularly if mast years occur consecutively, absence of abundant sources of more preferred foods. may drive such possum population increases. Indeed, Although moderate preference indices were possum trap-catch rates in the study area were stable recorded for red and mountain beech (foliage and bark between 1995 and 1999 but nearly doubled between combined), this feeding is unlikely to have a significant 1999 and 2001(A. Grant, unpubl. data). The effect of impact on beech canopies at the population level beech mast events on possum fecundity and juvenile because of the vast annual production of foliage and survival warrant investigation. relatively low possum densities in the study area. Using the foliage production data from Sweetapple Beech foliage and bark in possum diet and Fraser (1992) and food consumption data from When beech flowers and seed were absent during the Fitzgerald et al. (1981), possums ate just 0.3% of non-mast year, possums ate large quantities of beech annual beech foliage production in the current study foliage and bark. These items have previously been area. However, heavy impact on individual trees is shown to be virtually absent from possum diet in other possible if possum browsing was concentrated on a beech forests. They made up just 1.5% of annual diet few individuals. in a mixed silver beech-hardwood forest in South Bark chewing could potentially be far more Westland (Owen and Norton, 1995), and were damaging than the modest quantities involved would unrecorded in the diet of possums in two other mixed suggest. Ring barking could lead to death of small- to beech forests in Westland (Cochrane et al., 2003; moderate-sized branches, particularly where possums Sweetapple et al., 2004). While bark of non-beech were most abundant along the lower forest margin. species is sometimes taken as food (Kean, 1967), it has Indeed, heavy possum browsing damage to mountain not been recorded as exceeding 1% of total diet in beech was observed by the author along a valley-floor previous possum diet studies (Nugent et al., 2000). forest margin near the Mounds of Misery (Arthur’s The consumption of beech bark in this study is Pass National Park, South Island, New Zealand). In even more surprising given that bark from Nothofagus October 2001 c. 50% of the trees on this forest margin species has a high tannin (antifeedant secondary along a 1-km front exhibited some recent crown dieback metabolites) content, to the extent that it provided the (dead leaves still attached). Possums had stripped source of tannin for a leather-tanning industry in New much of the bark from all dead branches, with branches Zealand in the late 19th century (Cooper and Cambie, up to 5 cm in diameter affected. Dieback exceeded 1991). Possums may have eaten bark for the simple 50% of the crown in a few individuals, but was not sugar content of the inner bark (phloem). The extensive apparent on trees more than a few metres back from the bark consumption in this study may be a consequence forest margin. The dieback episode appeared to have of the limited range and quantity of preferred foods been of short duration as all affected branches had available to possums in non-mast years within the been dead for a similar length of time. Bark chewing

Table 5. Mean percent dry weight of beech foods in possum diet from central and eastern South Island, New Zealand. The number of possum stomachs analysed in each sample is given (n). ______Area/Forest type n Foliage Bark Seed Flowers Total ______Arthur’s Pass National Park - mountain beech forest1 4690009 Kaikoura - mountain beech/Kunzea forest2 10113023036 Craigieburn Forest Park - mountain beech forest3 27 < 1 3 24 2 29 Maruia Valley - red/silver beech forest1 30 < 1 0 75 0 75 Tutaki Valley - red/silver beech forest4 27 3 5 35 0 43 Nelson Lakes National Park - red/silver beech forest4 9 0 040040 ______1A. Byrom and P. Sweetapple unpubl. data 2G. Nugent and P. Sweetapple unpubl. data 3P. Sweetapple unpubl. data 4A. Byrom, Landcare Research, Lincoln, unpubl. data SWEETAPPLE: POSSUM DIET IN BEECH FOREST 165

by possums on beech trees has also been reported to (Coleman et al., 1985), with, as in this study, some cause death of branches in the North Island’s Ruahine possums travelling over 1 km through forest to feed on Range (Grant, 1956). Possum impacts on beech crowns pasture (Green and Coleman, 1986). at high-use sites, such as forest-pasture margins in Most of the few invertebrates eaten by possums in simple eastern and central beech forest warrant this study were also obtained while possums foraged investigation. on the ground. The lepidopteran larvae that dominated Recent unpublished studies of possum diet in this food category were usually encountered within central and eastern South Island indicate that high valley-floor grassland food layers within the possum levels of beech-food consumption are not unique to the stomachs. Owen and Norton (1995) reported that Hurunui study site. Possum diet was measured at six dipteran larvae made up 28% of the winter diet of sites with beech forest during 2000–2001 and foods possums from a mixed silver beech-hardwood forest in from beech species contributed 9–75% to total diet at South Westland, but dipteran larvae were not recorded all sites at the times the samples were taken (Table 5). in the current study or other studies of possum diet in beech forests. Ground-level feeding The ability of possums to utilise large quantities of low Potential possum impacts quality foods is not surprising given that their natural Considerable possum impacts in beech forests have habitats include sclerophyl Eucalyptus forests in been recorded previously (James, 1974; Rose et al., Australia, where their diet is dominated by low quality, 1993; Pekelharing et al., 1998; Sweetapple et al., highly defended Eucalyptus foliage (Freeland and Winter, 2004). These predominantly involve the collapse of 1975). However, there appear to be limits to how much seral forests, particularly those dominated by fuchsia, Eucalyptus foliage can be eaten: Possums supplement and the dieback of some other non-beech species their diet by foraging for herbs on the forest floor, even including southern rata (Metrosideros umbellata), though this means increased exposure to predators and kamahi (Weinmannia racemosa) and Hall’s totara increased foraging effort (Freeland and Winter, 1975). (Podocarpus hallii) (Rose et al., 1993; Sweetapple et This feeding strategy parallels that seen in the North al., 2004). Branch Hurunui during the non-mast year. Similar patterns of possum impact are likely in the The importance of forest floor fungi is another Hurunui Catchment. Beech species account for 82% notable feature of Hurunui possum diet. Fungi are of the forest biomass within the study area and these often recorded in possum diet (Nugent et al., 2000), species are unlikely to be significantly affected by but are usually only a minor component of total diet possum browsing, at least at the population level. (< 1%). In the North Branch Hurunui most possums Possum impacts on less common plants are likely to be had eaten some fungi, and it contributed 8.6–18.3% to significant. Woody seral species are heavily targeted annual diet. Given the scattered distribution of fungi by possums in the study area. Wineberry, ribbonwood within forest, and the large quantities of fungi and fuchsia have some of the highest preference or sometimes seen in stomachs, possums clearly actively utilisation indices in this study. Other plants with high foraged for fungi at this site. It is likely that this preference or at least high local utilisation indices are foraging behaviour reflects an increased relative mistletoe, pokaka, Polystichum vestitum, Blechnum preference for fungi due to the paucity of other preferred- procerum/B. penna-marina, forest-floor herbs, food sources in this low-diversity forest, because haumakaroa and the lianes Rubus cissoides and R. although fungi are common in all beech forests (P. schmidelioides. These plants would be the most useful Johnston, Landcare Research, Auckland, N.Z., pers. candidates for monitoring programmes aimed at comm.), they constitute < 1% of total diet in the more assessing possum impacts in this forest. Utilisation of floristically diverse mixed beech forests in Westland some preferred plants varied enormously between (Owen and Norton, 1995; Cochrane et al., 2003; sites within this study area (Table 4), therefore, possum Sweetapple et al., 2004). The consumption of fungi in impacts will be variable in space and any monitoring all areas may be due to their relatively high programme would have to be extensive, or targeted to concentrations of most macronutrients and low carbon- the sites of highest utilisation for the species of interest. to-nitrogen ratios (Clinton et al., 1999), which probably Possum impacts on mistletoe at this site appear to make them easily digested, rich sources of nutrients. be greatest during beech mast events. Consumption of The other foods eaten by ground-foraging possums mistletoe was significantly greater in the mast year were herbs, both from within and outside the limits of compared with the non-mast year. This may be a tall forest, and ferns such as Blechnum species and consequence of more extensive foraging in beech Polystichum vestitum. Similar species and quantities canopies by possums during mast years, particularly in of herbaceous ground foods have been recorded in the mountain beech (Table 2), the main host species for diet of possums from forest adjacent to pasture Alepis flavida and Peraxilla tetrapetela. A concurrent, 166 NEW ZEALAND JOURNAL OF ECOLOGY, VOL. 27, NO. 2, 2003

marked decline in mistletoe health was recorded during of Ecology 8: 21-35. the mast year in the study area, but not in a nearby area Cooper, R.C.; Cambie, R.C. 1991. New Zealand’s where possums had been controlled (A. Grant, unpubl. economic native plants. Oxford University Press, data). More intensive possum control may therefore Auckland, N.Z. be required during mast years compared with non-mast Cowan, P.E. 1990a. Effects of possums on fruiting of years to protect beech mistletoes in some areas. nikau. New Zealand Journal of Botany 29: 71-73. Cowan, P.E. 1990b. Brushtail possum. In: King, C.M. (Editor), The handbook of New Zealand mammals, Acknowledgements pp. 68-98. Oxford University Press, Auckland, N.Z. Staff from the Department of Conservation, Canterbury, Cowan, P.E.; Waddington, D.C. 1990. Suppression of collected the stomach samples. A. Byrom, P. Cowan, fruit production of the endemic forest tree, J. Parkes, W. Ruscoe, N. Poutu, and two anonymous Elaeocarpus dentatus, by introduced marsupial referees made useful comment on the draft manuscript. brushtail possums, Trichosurus vulpecula. New A. Wanrooy drafted the figures; C. Bezar edited and Zealand Journal of Botany 28: 217-224. W. Weller formatted the text. Data on vegetation Cowan, P.E.; Chilvers, B.L.; Efford, M.G.; McElrea, composition within the study area was drawn from the G.J. 1997. Effects of possum browsing on northern National Vegetation Survey databank (NVS) with the rata, Orongorongo Valley, Wellington, New help of C. Newell and L. Burrows. This work was Zealand. Journal of the Royal Society of New funded by the New Zealand Department of Zealand 27: 173-179. Conservation, Canterbury, under contract DOC1236. Elliott, P.G. 1996. Productivity and mortality of mohua (Mohoua ochrocephala). New Zealand Journal of Zoology 23: 229-237. References Elliott, P.G.; Dilks, P.J.; O’Donnell, C.F.J. 1996. The ecology of yellow crowned parakeets Allen, R.B. 1992. RECCE - an inventory method for (Cyanoramphus auriceps) in Nothofagus forest in describing New Zealand vegetation. Forest Fiordland, New Zealand. New Zealand Journal of Research Institute Bulletin No. 181. Forest Zoology 23: 249-265. Research Institute, Christchurch, N.Z. Fitzgerald, A.E.; Clarke, R.T.J.; Reid, C.S.W.; Alley, J.C.; Berben, P.H.; Dugdale, J.S.; Fitzgerald, Charleston, W.A.G.; Tarttelin, M.F.; Wyburn, B.M.; Knightbridge, P.I.; Meads, M.J.; Webster, R.S. 1981. Physical and nutritional characteristics R.A. 2001. Responses of litter-dwelling arthropods of the possum (Trichosurus vulpecula) in captivity. and house mice to beech seeding in the New Zealand Journal of Zoology 8: 551-562. Orongorongo Valley, New Zealand. Journal of Fitzgerald, B.M.; Daniel, M.J.; Fitzgerald, A.E.; Karl, the Royal Society of New Zealand 31: 425-452. B.J.; Meads, M.J.; Notman, P.R. 1996. Factors Bell, B.D. 1981. Breeding and condition of possums affecting the numbers of house mouse (Mus Trichosurus vulpecula in the Orongorongo Valley, musculus) in hard beech (Nothofagus truncata) near Wellington, New Zealand, 1966-1975. In: forest. Journal of the Royal Society of New Bell, B.D. (Editor), Proceedings of the first Zealand 26: 237-249. symposium on marsupials in New Zealand, pp. Freeland, W.J.; Winter, J.W. 1975. Evolutionary 87-139. Zoology Publications from Victoria consequences of eating: Trichosurus vulpecula University of Wellington No. 74. Victoria (Marsupialia) and the genus Eucalyptus. Journal University, Wellington, N.Z. of Chemical Ecology 1: 439-455. Clinton, P.W.; Buchanan, P.K.; Allen, R.B. 1999. Grant, P.J. 1956. Opossum damage in beech forests Nutrient composition of epigeous fungal Ruahine Range, Hawke’s Bay. New Zealand sporocarps growing on different substrates in a Journal of Forestry 7: 111-113. New Zealand mountain beech forest. New Zealand Green, W.Q.; Coleman, J.D. 1986. Movement of Journal of Botany 37: 149-153. possums (Trichosurus vulpecula) between forest Cochrane, C.H.; Norton, D.A.; Miller, C.J.; Allen, and pasture in Westland, New Zealand: R.B. 2003. Brushtail possum (Trichosurus implications for bovine tuberculosis transmission. vulpecula) diet in a north Westland mixed-beech New Zealand Journal of Ecology 9: 57-69. (Nothofagus) forest. New Zealand Journal of James, I.L. 1974. Mammals and beech (Nothofagus) Ecology 27: 61-65. forests. Proceedings of the New Zealand Coleman, J.D.; Green, W.Q.; Polson, J.G. 1985. Diet Ecological Society 21: 41-44. of brushtail possums over a pasture–alpine gradient Kean, R.I. 1967. Behaviour and territorialism in in Westland, New Zealand. New Zealand Journal Trichosurus vulpecula (Marsupialia). Proceedings SWEETAPPLE: POSSUM DIET IN BEECH FOREST 167

of the New Zealand Ecological Society 14: 71-78. Pekelharing, C.J.; Frampton, C.M.; Suisted, P.A. 1998. King, C.M. 1982. Age structure and reproduction in Seasonal variation in the impacts of brushtailed feral New Zealand populations of house mouse possums (Trichosurus vulpecula) on five palatable (Mus musculus), in relation to seedfall of southern plant species in New Zealand beech (Nothofagus) beech. New Zealand Journal of Zoology 9: 467- forest. New Zealand Journal of Ecology 22: 141- 480. 148. King, C.M. 1983. The relationships between beech Rose, A.B.; Pekelharing, C.J.; Platt, K.H.; Woolmore, (Nothofagus sp.) seedfall and populations of mice C.B. 1993. Impact of invading brushtail possum (Mus musculus), and the demographic and dietary populations on mixed beech-broadleaved forests, responses of stoats (Mustela erminea), in three south Westland, New Zealand. New Zealand New Zealand forests. Journal of Animal Ecology Journal of Ecology 17: 19-28. 52: 141–166. Sadleir, R. 2000. Evidence of possums as predators of Loehle, C.; Rittenhouse, L.R. 1982. Analysis of forage native animals. In: Montague, T.L. (Editor), The preference indices. Journal of Range Management brushtail possum: biology, impact and 35: 316-319. management of an introduced marsupial. pp. Mitchell, A.D.; Froden, D.G.; Heads, M.J. 1997. 126-131. Manaaki Whenua Press, Lincoln, N.Z. Reinstatement of Raukaua, a genus of the Spurr, E.B.; Warburton, B. 1991. Methods of measuring centred in New Zealand. New Zealand the proportions of plant species present in forest Journal of Botany 35: 309-315. and their effect on estimates of bird preferences Murphy, E.C. 1992. The effects of a natural increase for plant species. New Zealand Journal of Ecology in food supply on a wild population of house mice. 15: 171-175. New Zealand Journal of Ecology 16: 33-40. Sweetapple, P.J.; Fraser, K.W. 1992; Litterfall from a Murphy, E.C.; Dowding, J.E. 1995. Ecology of the mixed red beech (Nothofagus fusca)-silver beech stoat in Nothofagus forest: home range, habitat (Nothofagus menziesii) forest, central North Island, use and diet at different stages of the beech mast New Zealand. New Zealand Journal of Botany cycle. New Zealand Journal of Ecology 19: 97- 30: 263-269 109. Sweetapple, P.J.; Nugent, G. 1998. Comparison of Nugent, G.; Sweetapple, P.J.; Coleman, J.D.; Suisted, two techniques for assessing possum (Trichosurus P. 2000. Possum feeding patterns: dietary tactics vulpecula) diet from stomach contents. New of a reluctant folivore. In: Montague, T. (Editor), Zealand Journal of Ecology 22: 181-188. The brushtail possum: biology, impact and Sweetapple, P.J.; Fraser, K.W.; Knightbridge, P.I.; management of an introduced marsupial, pp. 10- 2004; Diet and impacts of brushtail possum 23. Manaaki Whenua Press, Lincoln, N.Z. populations across an invasion front in South O’Donnell, C.F.J.; Phillipson, S.M. 1996. Predicting Westland, New Zealand. New Zealand Journal of the incidence of mohua predation from the seedfall, Ecology 28: (in press). mouse and predator fluctuations in beech forests. Ward, G.D. 1978. Habitat use and home-range of New Zealand Journal of Zoology 23: 287-293. radio-tagged opossums Trichosurus vulpecula Owen, H.J.; Norton, D.A. 1995. The diet of introduced (Kerr) in New Zealand lowland forest. In: brushtail possums Trichosurus vulpecula in a Montgomery, G.G. (Editor), The ecology of low-diversity New Zealand Nothofagus forest and arboreal folivores, pp. 267-287. Smithsonian possible implications for conservation Institution Press, Washington, D.C., U.S.A. management. Biological Conservation 71: 339- Wardle, J.A. 1984. The New Zealand : ecology, 345. utilisation and management. New Zealand Forest Payton, I. 2000. Damage to native forests. In: Service, Christchurch, N.Z. Montague, T.L. (Editor), The brushtail possum: Wilson, P.R.; Karl, B.J.; Toft, R.J.; Beggs, J.R.; Taylor, biology, impact and management of an introduced R.H. 1998. The role of introduced predators and marsupial, pp. 111-125. Manaaki Whenua Press, competitors in the decline of kaka (Nestor Lincoln, N.Z. meridionalis) populations in New Zealand. Biological Conservation 83: 175-185.

Editorial Board member: Graham Hickling 168 NEW ZEALAND JOURNAL OF ECOLOGY, VOL. 27, NO. 2, 2003